Newer
Older
4001
4002
4003
4004
4005
4006
4007
4008
4009
4010
4011
4012
4013
4014
4015
4016
4017
4018
4019
4020
4021
4022
4023
4024
4025
4026
4027
4028
4029
4030
4031
4032
4033
4034
4035
4036
4037
4038
4039
4040
4041
4042
4043
4044
4045
4046
4047
4048
4049
4050
4051
4052
4053
4054
4055
4056
4057
4058
4059
4060
4061
4062
4063
4064
4065
4066
4067
4068
4069
4070
4071
4072
4073
4074
4075
4076
4077
4078
4079
4080
4081
4082
4083
4084
4085
4086
4087
4088
4089
4090
4091
4092
4093
4094
4095
4096
4097
4098
4099
4100
4101
4102
4103
4104
4105
4106
4107
4108
4109
4110
4111
4112
4113
4114
4115
4116
4117
4118
4119
4120
4121
4122
4123
4124
4125
4126
4127
4128
4129
4130
4131
4132
4133
4134
4135
4136
4137
4138
4139
4140
4141
4142
4143
4144
4145
4146
4147
4148
4149
4150
4151
4152
4153
4154
4155
4156
4157
4158
4159
4160
4161
4162
4163
4164
4165
4166
4167
4168
4169
4170
4171
4172
4173
4174
4175
4176
4177
4178
4179
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
4208
4209
4210
4211
4212
4213
4214
4215
4216
4217
4218
4219
4220
4221
4222
4223
4224
4225
4226
4227
4228
4229
4230
4231
4232
4233
4234
4235
4236
4237
4238
4239
4240
4241
4242
case NN_lidt: // Load Interrupt Descriptor Table Register
case NN_lgs: // Load Full Pointer to GS:xx
case NN_lss: // Load Full Pointer to SS:xx
case NN_lds: // Load Full Pointer to DS:xx
case NN_les: // Load Full Pointer to ES:xx
case NN_lfs: // Load Full Pointer to FS:xx
case NN_lldt: // Load Local Descriptor Table Register
case NN_lmsw: // Load Machine Status Word
case NN_lock: // Assert LOCK# Signal Prefix
case NN_lods: // Load String
return false;
break;
case NN_loopw: // Loop while ECX != 0
case NN_loop: // Loop while CX != 0
case NN_loopd: // Loop while ECX != 0
case NN_loopq: // Loop while RCX != 0
case NN_loopwe: // Loop while CX != 0 and ZF=1
case NN_loope: // Loop while rCX != 0 and ZF=1
case NN_loopde: // Loop while ECX != 0 and ZF=1
case NN_loopqe: // Loop while RCX != 0 and ZF=1
case NN_loopwne: // Loop while CX != 0 and ZF=0
case NN_loopne: // Loop while rCX != 0 and ZF=0
case NN_loopdne: // Loop while ECX != 0 and ZF=0
case NN_loopqne: // Loop while RCX != 0 and ZF=0
return false;
break;
case NN_lsl: // Load Segment Limit
case NN_ltr: // Load Task Register
return false;
break;
case NN_mov: // Move Data
case NN_movsp: // Move to/from Special Registers
case NN_movs: // Move Byte(s) from String to String
return this->BuildMoveRTL(SMP_NULL_OPERATOR);
case NN_movsx: // Move with Sign-Extend
return this->BuildUnary2OpndRTL(SMP_SIGN_EXTEND);
case NN_movzx: // Move with Zero-Extend
return this->BuildUnary2OpndRTL(SMP_ZERO_EXTEND);
case NN_mul: // Unsigned Multiplication of AL or AX
return this->BuildMultiplyDivideRTL(SMP_U_MULTIPLY);
case NN_neg: // Two's Complement Negation
return this->BuildUnaryRTL(SMP_NEGATE);
case NN_nop: // No Operation
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
case NN_not: // One's Complement Negation
return this->BuildUnaryRTL(SMP_BITWISE_NOT);
case NN_or: // Logical Inclusive OR
return this->BuildBinaryRTL(SMP_BITWISE_OR);
case NN_out: // Output to Port
return this->BuildBinaryRTL(SMP_OUTPUT);
case NN_outs: // Output Byte(s) to Port
return false;
break;
case NN_pop: // Pop a word from the Stack
case NN_popaw: // Pop all General Registers
case NN_popa: // Pop all General Registers
case NN_popad: // Pop all General Registers (use32)
case NN_popaq: // Pop all General Registers (use64)
case NN_popfw: // Pop Stack into Flags Register
case NN_popf: // Pop Stack into Flags Register
case NN_popfd: // Pop Stack into Eflags Register
case NN_popfq: // Pop Stack into Rflags Register
return this->BuildPopRTL();
case NN_push: // Push Operand onto the Stack
case NN_pushaw: // Push all General Registers
case NN_pusha: // Push all General Registers
case NN_pushad: // Push all General Registers (use32)
case NN_pushaq: // Push all General Registers (use64)
case NN_pushfw: // Push Flags Register onto the Stack
case NN_pushf: // Push Flags Register onto the Stack
case NN_pushfd: // Push Flags Register onto the Stack (use32)
case NN_pushfq: // Push Flags Register onto the Stack (use64)
return this->BuildPushRTL();
case NN_rcl: // Rotate Through Carry Left
return this->BuildBinaryPlusFlagsRTL(SMP_ROTATE_LEFT_CARRY);
case NN_rcr: // Rotate Through Carry Right
return this->BuildBinaryPlusFlagsRTL(SMP_ROTATE_RIGHT_CARRY);
case NN_rol: // Rotate Left
return this->BuildBinaryRTL(SMP_ROTATE_LEFT);
case NN_ror: // Rotate Right
return this->BuildBinaryRTL(SMP_ROTATE_RIGHT);
case NN_rep: // Repeat String Operation
case NN_repe: // Repeat String Operation while ZF=1
case NN_repne: // Repeat String Operation while ZF=0
return false;
break;
case NN_retn: // Return Near from Procedure
case NN_retf: // Return Far from Procedure
return this->BuildReturnRTL();
case NN_sahf: // Store AH into Flags Register
return this->BuildMoveRTL(SMP_NULL_OPERATOR);
case NN_sal: // Shift Arithmetic Left
return this->BuildBinaryRTL(SMP_S_LEFT_SHIFT);
case NN_sar: // Shift Arithmetic Right
return this->BuildBinaryRTL(SMP_S_RIGHT_SHIFT);
case NN_shl: // Shift Logical Left
return this->BuildBinaryRTL(SMP_U_LEFT_SHIFT);
case NN_shr: // Shift Logical Right
return this->BuildBinaryRTL(SMP_U_RIGHT_SHIFT);
case NN_sbb: // Integer Subtraction with Borrow
return this->BuildBinaryPlusFlagsRTL(SMP_SUBTRACT_BORROW);
case NN_scas: // Compare String
return this->BuildBinaryPlusFlagsRTL(SMP_U_COMPARE);
case NN_seta: // Set Byte if Above (CF=0 & ZF=0)
case NN_setae: // Set Byte if Above or Equal (CF=0)
case NN_setb: // Set Byte if Below (CF=1)
case NN_setbe: // Set Byte if Below or Equal (CF=1 | ZF=1)
case NN_setc: // Set Byte if Carry (CF=1)
case NN_sete: // Set Byte if Equal (ZF=1)
case NN_setg: // Set Byte if Greater (ZF=0 & SF=OF)
case NN_setge: // Set Byte if Greater or Equal (SF=OF)
case NN_setl: // Set Byte if Less (SF!=OF)
case NN_setle: // Set Byte if Less or Equal (ZF=1 | SF!=OF)
case NN_setna: // Set Byte if Not Above (CF=1 | ZF=1)
case NN_setnae: // Set Byte if Not Above or Equal (CF=1)
case NN_setnb: // Set Byte if Not Below (CF=0)
case NN_setnbe: // Set Byte if Not Below or Equal (CF=0 & ZF=0)
case NN_setnc: // Set Byte if Not Carry (CF=0)
case NN_setne: // Set Byte if Not Equal (ZF=0)
case NN_setng: // Set Byte if Not Greater (ZF=1 | SF!=OF)
case NN_setnge: // Set Byte if Not Greater or Equal (ZF=1)
case NN_setnl: // Set Byte if Not Less (SF=OF)
case NN_setnle: // Set Byte if Not Less or Equal (ZF=0 & SF=OF)
case NN_setno: // Set Byte if Not Overflow (OF=0)
case NN_setnp: // Set Byte if Not Parity (PF=0)
case NN_setns: // Set Byte if Not Sign (SF=0)
case NN_setnz: // Set Byte if Not Zero (ZF=0)
case NN_seto: // Set Byte if Overflow (OF=1)
case NN_setp: // Set Byte if Parity (PF=1)
case NN_setpe: // Set Byte if Parity Even (PF=1)
case NN_setpo: // Set Byte if Parity Odd (PF=0)
case NN_sets: // Set Byte if Sign (SF=1)
case NN_setz: // Set Byte if Zero (ZF=1)
// Destination always get set to NUMERIC 0 or 1, depending on
// the condition and the relevant flags bits. Best way to model
// this in an RTL is to perform an unspecified unary NUMERIC
// operation on the flags register and assign the result to the
// destination operand, making it always NUMERIC.
return this->BuildUnary2OpndRTL(SMP_UNARY_NUMERIC_OPERATION);
case NN_sgdt: // Store Global Descriptor Table Register
case NN_sidt: // Store Interrupt Descriptor Table Register
return false;
break;
case NN_shld: // Double Precision Shift Left
return this->BuildDoubleShiftRTL(SMP_U_LEFT_SHIFT);
case NN_shrd: // Double Precision Shift Right
return this->BuildDoubleShiftRTL(SMP_U_RIGHT_SHIFT);
case NN_sldt: // Store Local Descriptor Table Register
case NN_smsw: // Store Machine Status Word
return false;
break;
case NN_stc: // Set Carry Flag
case NN_std: // Set Direction Flag
return this->BuildUnaryRTL(SMP_UNARY_NUMERIC_OPERATION);
case NN_sti: // Set Interrupt Flag
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
case NN_stos: // Store String
return this->BuildMoveRTL(SMP_NULL_OPERATOR);
case NN_str: // Store Task Register
return false;
break;
case NN_sub: // Integer Subtraction
return this->BuildBinaryRTL(SMP_SUBTRACT);
case NN_test: // Logical Compare
return this->BuildFlagsDestBinaryRTL(SMP_U_COMPARE);
case NN_verr: // Verify a Segment for Reading
case NN_verw: // Verify a Segment for Writing
case NN_wait: // Wait until BUSY# Pin is Inactive (HIGH)
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
if (NN_wait != this->SMPcmd.itype)
this->RTL.ExtraKills.push_back(FlagsOp);
return true;
case NN_xchg: // Exchange Register/Memory with Register
return this->BuildExchangeRTL();
case NN_xlat: // Table Lookup Translation
return false;
break;
case NN_xor: // Logical Exclusive OR
return this->BuildBinaryRTL(SMP_BITWISE_XOR);
//
// 486 instructions
//
case NN_cmpxchg: // Compare and Exchange
return this->BuildCompareExchangeRTL();
case NN_bswap: // Swap bits in EAX
return this->BuildUnaryRTL(SMP_UNARY_NUMERIC_OPERATION);
4244
4245
4246
4247
4248
4249
4250
4251
4252
4253
4254
4255
4256
4257
4258
4259
4260
4261
4262
4263
4264
4265
4266
4267
4268
4269
4270
4271
4272
4273
4274
4275
4276
4277
4278
4279
4280
4281
4282
4283
4284
4285
4286
4287
4288
4289
4290
4291
4292
4293
4294
4295
4296
4297
4298
4299
4300
4301
4302
4303
4304
4305
4306
4307
4308
4309
4310
4311
4312
4313
4314
4315
4316
4317
4318
4319
4320
4321
4322
4323
4324
4325
4326
4327
4328
4329
4330
4331
4332
4333
4334
4335
4336
4337
4338
4339
4340
4341
4342
4343
4344
4345
4346
4347
4348
4349
4350
4351
4352
4353
4354
4355
4356
4357
4358
4359
4360
4361
4362
4363
4364
4365
4366
4367
4368
4369
4370
4371
4372
4373
4374
4375
4376
4377
4378
4379
4380
4381
4382
4383
4384
4385
4386
4387
4388
4389
4390
4391
4392
4393
4394
4395
4396
4397
4398
4399
4400
4401
4402
4403
4404
4405
4406
4407
4408
4409
4410
4411
4412
4413
4414
4415
4416
4417
4418
4419
4420
4421
4422
4423
4424
4425
4426
4427
4428
4429
4430
4431
4432
4433
4434
case NN_xadd: // t<-dest; dest<-src+dest; src<-t
return this->BuildExchangeAddRTL();
case NN_invd: // Invalidate Data Cache
case NN_wbinvd: // Invalidate Data Cache (write changes)
case NN_invlpg: // Invalidate TLB entry
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
//
// Pentium instructions
//
case NN_rdmsr: // Read Machine Status Register
return this->BuildOptType8RTL();
case NN_wrmsr: // Write Machine Status Register
return false;
break;
case NN_cpuid: // Get CPU ID
return this->BuildOptType8RTL();
case NN_cmpxchg8b: // Compare and Exchange Eight Bytes
return false;
break;
case NN_rdtsc: // Read Time Stamp Counter
return this->BuildOptType8RTL();
case NN_rsm: // Resume from System Management Mode
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
//
// Pentium Pro instructions
//
case NN_cmova: // Move if Above (CF=0 & ZF=0)
case NN_cmovb: // Move if Below (CF=1)
case NN_cmovbe: // Move if Below or Equal (CF=1 | ZF=1)
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_cmovg: // Move if Greater (ZF=0 & SF=OF)
return this->BuildMoveRTL(SMP_GREATER_THAN);
case NN_cmovge: // Move if Greater or Equal (SF=OF)
return this->BuildMoveRTL(SMP_GREATER_EQUAL);
case NN_cmovl: // Move if Less (SF!=OF)
return this->BuildMoveRTL(SMP_LESS_THAN);
case NN_cmovle: // Move if Less or Equal (ZF=1 | SF!=OF)
return this->BuildMoveRTL(SMP_LESS_EQUAL);
case NN_cmovnb: // Move if Not Below (CF=0)
case NN_cmovno: // Move if Not Overflow (OF=0)
case NN_cmovnp: // Move if Not Parity (PF=0)
case NN_cmovns: // Move if Not Sign (SF=0)
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_cmovnz: // Move if Not Zero (ZF=0)
return this->BuildMoveRTL(SMP_NOT_EQUAL);
case NN_cmovo: // Move if Overflow (OF=1)
case NN_cmovp: // Move if Parity (PF=1)
case NN_cmovs: // Move if Sign (SF=1)
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_cmovz: // Move if Zero (ZF=1)
return this->BuildMoveRTL(SMP_EQUAL);
case NN_fcmovb: // Floating Move if Below
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_fcmove: // Floating Move if Equal
return this->BuildMoveRTL(SMP_EQUAL);
case NN_fcmovbe: // Floating Move if Below or Equal
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_fcmovu: // Floating Move if Unordered
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_fcmovnb: // Floating Move if Not Below
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_fcmovne: // Floating Move if Not Equal
return this->BuildMoveRTL(SMP_NOT_EQUAL);
case NN_fcmovnbe: // Floating Move if Not Below or Equal
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_fcmovnu: // Floating Move if Not Unordered
return this->BuildMoveRTL(SMP_BINARY_NUMERIC_OPERATION);
case NN_fcomi: // FP Compare: result in EFLAGS
case NN_fucomi: // FP Unordered Compare: result in EFLAGS
case NN_fcomip: // FP Compare: result in EFLAGS: pop stack
case NN_fucomip: // FP Unordered Compare: result in EFLAGS: pop stack
return false;
break;
case NN_rdpmc: // Read Performance Monitor Counter
return this->BuildOptType8RTL();
//
// FPP instructions
//
case NN_fld: // Load Real
case NN_fst: // Store Real
case NN_fstp: // Store Real and Pop
return this->BuildMoveRTL(SMP_NULL_OPERATOR);
case NN_fxch: // Exchange Registers
// FP registers remain NUMERIC anyway, so this is a no-op to our type system.
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
case NN_fild: // Load Integer
case NN_fist: // Store Integer
case NN_fistp: // Store Integer and Pop
case NN_fbld: // Load BCD
case NN_fbstp: // Store BCD and Pop
return this->BuildMoveRTL(SMP_NULL_OPERATOR);
case NN_fadd: // Add Real
case NN_faddp: // Add Real and Pop
case NN_fiadd: // Add Integer
case NN_fsub: // Subtract Real
case NN_fsubp: // Subtract Real and Pop
case NN_fisub: // Subtract Integer
case NN_fsubr: // Subtract Real Reversed
case NN_fsubrp: // Subtract Real Reversed and Pop
case NN_fisubr: // Subtract Integer Reversed
case NN_fmul: // Multiply Real
case NN_fmulp: // Multiply Real and Pop
case NN_fimul: // Multiply Integer
case NN_fdiv: // Divide Real
case NN_fdivp: // Divide Real and Pop
case NN_fidiv: // Divide Integer
case NN_fdivr: // Divide Real Reversed
case NN_fdivrp: // Divide Real Reversed and Pop
case NN_fidivr: // Divide Integer Reversed
return this->BuildBinaryRTL(SMP_BINARY_FLOATING_ARITHMETIC);
case NN_fsqrt: // Square Root
case NN_fscale: // Scale: st(0) <- st(0) * 2^st(1)
case NN_fprem: // Partial Remainder
case NN_frndint: // Round to Integer
case NN_fxtract: // Extract exponent and significand
case NN_fabs: // Absolute value
case NN_fchs: // Change Sign
return this->BuildUnaryRTL(SMP_UNARY_FLOATING_ARITHMETIC);
case NN_fcom: // Compare Real
case NN_fcomp: // Compare Real and Pop
case NN_fcompp: // Compare Real and Pop Twice
case NN_ficom: // Compare Integer
case NN_ficomp: // Compare Integer and Pop
case NN_ftst: // Test
case NN_fxam: // Examine
// Floating comparison instructions use FP reg stack locations
// as sources and set only the FP flags. All of these are numeric
// type and we don't track any of them, so all such instructions
// can be considered to be no-ops.
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
case NN_fptan: // Partial tangent
case NN_fpatan: // Partial arctangent
case NN_f2xm1: // 2^x - 1
case NN_fyl2x: // Y * lg2(X)
case NN_fyl2xp1: // Y * lg2(X+1)
// We can consider it a unary operation when both arguments come
// off the floating point register stack, unless we ever start
4436
4437
4438
4439
4440
4441
4442
4443
4444
4445
4446
4447
4448
4449
4450
4451
4452
4453
4454
4455
4456
4457
4458
4459
4460
4461
4462
4463
4464
4465
4466
4467
4468
4469
4470
4471
4472
4473
4474
4475
4476
4477
4478
4479
4480
4481
4482
4483
4484
4485
4486
4487
4488
4489
4490
4491
4492
4493
4494
4495
4496
4497
4498
4499
4500
4501
4502
4503
4504
4505
4506
4507
4508
4509
4510
4511
4512
4513
4514
4515
4516
4517
4518
4519
4520
4521
4522
4523
4524
4525
4526
4527
4528
4529
4530
4531
4532
4533
4534
4535
4536
4537
4538
4539
4540
4541
4542
4543
4544
4545
4546
4547
4548
4549
4550
4551
4552
4553
4554
4555
4556
4557
4558
4559
4560
4561
4562
4563
4564
4565
4566
4567
4568
4569
4570
4571
4572
4573
4574
4575
4576
4577
4578
4579
4580
4581
4582
4583
4584
4585
4586
4587
4588
4589
4590
4591
4592
4593
4594
4595
4596
4597
4598
4599
4600
4601
4602
4603
4604
4605
4606
4607
4608
4609
4610
4611
4612
4613
4614
4615
4616
4617
4618
4619
4620
4621
4622
4623
4624
4625
4626
4627
4628
4629
4630
4631
4632
4633
4634
4635
4636
4637
4638
4639
4640
4641
4642
4643
4644
4645
4646
4647
4648
4649
4650
4651
4652
4653
4654
4655
4656
4657
4658
4659
4660
4661
4662
4663
4664
4665
4666
4667
4668
4669
4670
4671
4672
4673
4674
4675
4676
4677
4678
4679
4680
4681
4682
4683
4684
4685
4686
4687
4688
4689
4690
4691
4692
4693
4694
4695
4696
4697
4698
4699
4700
4701
4702
4703
4704
4705
4706
4707
4708
4709
4710
4711
4712
4713
4714
4715
4716
4717
4718
4719
4720
4721
4722
4723
4724
4725
4726
4727
4728
4729
4730
4731
4732
4733
4734
4735
4736
4737
4738
4739
4740
4741
4742
4743
4744
4745
4746
4747
4748
4749
4750
4751
4752
4753
4754
4755
4756
4757
4758
4759
4760
4761
4762
4763
4764
4765
4766
4767
4768
4769
4770
4771
4772
4773
4774
4775
4776
4777
4778
4779
4780
4781
4782
4783
4784
4785
4786
4787
4788
4789
4790
4791
4792
4793
4794
4795
4796
4797
4798
4799
4800
4801
4802
4803
4804
4805
4806
4807
4808
4809
4810
4811
4812
4813
4814
4815
4816
4817
4818
4819
4820
4821
4822
4823
4824
4825
4826
4827
4828
4829
4830
4831
4832
4833
4834
4835
4836
4837
4838
4839
4840
4841
4842
4843
4844
4845
4846
4847
4848
4849
4850
4851
4852
4853
4854
4855
4856
4857
4858
4859
// modeling the different locations in the FP register stack.
return this->BuildUnaryRTL(SMP_UNARY_FLOATING_ARITHMETIC);
case NN_fldz: // Load +0.0
case NN_fld1: // Load +1.0
case NN_fldpi: // Load PI=3.14...
case NN_fldl2t: // Load lg2(10)
case NN_fldl2e: // Load lg2(e)
case NN_fldlg2: // Load lg10(2)
case NN_fldln2: // Load ln(2)
case NN_finit: // Initialize Processor
case NN_fninit: // Initialize Processor (no wait)
case NN_fsetpm: // Set Protected Mode
case NN_fldcw: // Load Control Word
case NN_fstcw: // Store Control Word
case NN_fnstcw: // Store Control Word (no wait)
case NN_fstsw: // Store Status Word
case NN_fnstsw: // Store Status Word (no wait)
case NN_fclex: // Clear Exceptions
case NN_fnclex: // Clear Exceptions (no wait)
// Floating point stack and control word and flags operations
// with no memory operands are no-ops to us.
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
case NN_fstenv: // Store Environment
case NN_fnstenv: // Store Environment (no wait)
case NN_fldenv: // Load Environment
case NN_fsave: // Save State
case NN_fnsave: // Save State (no wait)
case NN_frstor: // Restore State
case NN_fincstp: // Increment Stack Pointer
case NN_fdecstp: // Decrement Stack Pointer
case NN_ffree: // Free Register
case NN_fnop: // No Operation
case NN_feni: // (8087 only)
case NN_fneni: // (no wait) (8087 only)
case NN_fdisi: // (8087 only)
case NN_fndisi: // (no wait) (8087 only)
return false;
break;
//
// 80387 instructions
//
case NN_fprem1: // Partial Remainder ( < half )
case NN_fsincos: // t<-cos(st); st<-sin(st); push t
case NN_fsin: // Sine
case NN_fcos: // Cosine
case NN_fucom: // Compare Unordered Real
case NN_fucomp: // Compare Unordered Real and Pop
case NN_fucompp: // Compare Unordered Real and Pop Twice
// Floating point stack and control word and flags operations
// with no memory operands are no-ops to us.
NopRT = new SMPRegTransfer;
NopRT->SetOperator(SMP_NULL_OPERATOR);
this->RTL.push_back(NopRT);
NopRT = NULL;
return true;
//
// Instructions added 28.02.96
//
case NN_setalc: // Set AL to Carry Flag
case NN_svdc: // Save Register and Descriptor
case NN_rsdc: // Restore Register and Descriptor
case NN_svldt: // Save LDTR and Descriptor
case NN_rsldt: // Restore LDTR and Descriptor
case NN_svts: // Save TR and Descriptor
case NN_rsts: // Restore TR and Descriptor
case NN_icebp: // ICE Break Point
case NN_loadall: // Load the entire CPU state from ES:EDI
//
// MMX instructions
//
case NN_emms: // Empty MMX state
case NN_movd: // Move 32 bits
case NN_movq: // Move 64 bits
case NN_packsswb: // Pack with Signed Saturation (Word->Byte)
case NN_packssdw: // Pack with Signed Saturation (Dword->Word)
case NN_packuswb: // Pack with Unsigned Saturation (Word->Byte)
case NN_paddb: // Packed Add Byte
case NN_paddw: // Packed Add Word
case NN_paddd: // Packed Add Dword
case NN_paddsb: // Packed Add with Saturation (Byte)
case NN_paddsw: // Packed Add with Saturation (Word)
case NN_paddusb: // Packed Add Unsigned with Saturation (Byte)
case NN_paddusw: // Packed Add Unsigned with Saturation (Word)
case NN_pand: // Bitwise Logical And
case NN_pandn: // Bitwise Logical And Not
case NN_pcmpeqb: // Packed Compare for Equal (Byte)
case NN_pcmpeqw: // Packed Compare for Equal (Word)
case NN_pcmpeqd: // Packed Compare for Equal (Dword)
case NN_pcmpgtb: // Packed Compare for Greater Than (Byte)
case NN_pcmpgtw: // Packed Compare for Greater Than (Word)
case NN_pcmpgtd: // Packed Compare for Greater Than (Dword)
case NN_pmaddwd: // Packed Multiply and Add
case NN_pmulhw: // Packed Multiply High
case NN_pmullw: // Packed Multiply Low
case NN_por: // Bitwise Logical Or
case NN_psllw: // Packed Shift Left Logical (Word)
case NN_pslld: // Packed Shift Left Logical (Dword)
case NN_psllq: // Packed Shift Left Logical (Qword)
case NN_psraw: // Packed Shift Right Arithmetic (Word)
case NN_psrad: // Packed Shift Right Arithmetic (Dword)
case NN_psrlw: // Packed Shift Right Logical (Word)
case NN_psrld: // Packed Shift Right Logical (Dword)
case NN_psrlq: // Packed Shift Right Logical (Qword)
case NN_psubb: // Packed Subtract Byte
case NN_psubw: // Packed Subtract Word
case NN_psubd: // Packed Subtract Dword
case NN_psubsb: // Packed Subtract with Saturation (Byte)
case NN_psubsw: // Packed Subtract with Saturation (Word)
case NN_psubusb: // Packed Subtract Unsigned with Saturation (Byte)
case NN_psubusw: // Packed Subtract Unsigned with Saturation (Word)
case NN_punpckhbw: // Unpack High Packed Data (Byte->Word)
case NN_punpckhwd: // Unpack High Packed Data (Word->Dword)
case NN_punpckhdq: // Unpack High Packed Data (Dword->Qword)
case NN_punpcklbw: // Unpack Low Packed Data (Byte->Word)
case NN_punpcklwd: // Unpack Low Packed Data (Word->Dword)
case NN_punpckldq: // Unpack Low Packed Data (Dword->Qword)
case NN_pxor: // Bitwise Logical Exclusive Or
//
// Undocumented Deschutes processor instructions
//
case NN_fxsave: // Fast save FP context
case NN_fxrstor: // Fast restore FP context
// Pentium II instructions
case NN_sysenter: // Fast Transition to System Call Entry Point
case NN_sysexit: // Fast Transition from System Call Entry Point
// 3DNow! instructions
case NN_pavgusb: // Packed 8-bit Unsigned Integer Averaging
case NN_pfadd: // Packed Floating-Point Addition
case NN_pfsub: // Packed Floating-Point Subtraction
case NN_pfsubr: // Packed Floating-Point Reverse Subtraction
case NN_pfacc: // Packed Floating-Point Accumulate
case NN_pfcmpge: // Packed Floating-Point Comparison: Greater or Equal
case NN_pfcmpgt: // Packed Floating-Point Comparison: Greater
case NN_pfcmpeq: // Packed Floating-Point Comparison: Equal
case NN_pfmin: // Packed Floating-Point Minimum
case NN_pfmax: // Packed Floating-Point Maximum
case NN_pi2fd: // Packed 32-bit Integer to Floating-Point
case NN_pf2id: // Packed Floating-Point to 32-bit Integer
case NN_pfrcp: // Packed Floating-Point Reciprocal Approximation
case NN_pfrsqrt: // Packed Floating-Point Reciprocal Square Root Approximation
case NN_pfmul: // Packed Floating-Point Multiplication
case NN_pfrcpit1: // Packed Floating-Point Reciprocal First Iteration Step
case NN_pfrsqit1: // Packed Floating-Point Reciprocal Square Root First Iteration Step
case NN_pfrcpit2: // Packed Floating-Point Reciprocal Second Iteration Step
case NN_pmulhrw: // Packed Floating-Point 16-bit Integer Multiply with rounding
case NN_femms: // Faster entry/exit of the MMX or floating-point state
case NN_prefetch: // Prefetch at least a 32-byte line into L1 data cache
case NN_prefetchw: // Prefetch processor cache line into L1 data cache (mark as modified)
// Pentium III instructions
case NN_addps: // Packed Single-FP Add
case NN_addss: // Scalar Single-FP Add
case NN_andnps: // Bitwise Logical And Not for Single-FP
case NN_andps: // Bitwise Logical And for Single-FP
case NN_cmpps: // Packed Single-FP Compare
case NN_cmpss: // Scalar Single-FP Compare
case NN_comiss: // Scalar Ordered Single-FP Compare and Set EFLAGS
case NN_cvtpi2ps: // Packed signed INT32 to Packed Single-FP conversion
case NN_cvtps2pi: // Packed Single-FP to Packed INT32 conversion
case NN_cvtsi2ss: // Scalar signed INT32 to Single-FP conversion
case NN_cvtss2si: // Scalar Single-FP to signed INT32 conversion
case NN_cvttps2pi: // Packed Single-FP to Packed INT32 conversion (truncate)
case NN_cvttss2si: // Scalar Single-FP to signed INT32 conversion (truncate)
case NN_divps: // Packed Single-FP Divide
case NN_divss: // Scalar Single-FP Divide
case NN_ldmxcsr: // Load Streaming SIMD Extensions Technology Control/Status Register
case NN_maxps: // Packed Single-FP Maximum
case NN_maxss: // Scalar Single-FP Maximum
case NN_minps: // Packed Single-FP Minimum
case NN_minss: // Scalar Single-FP Minimum
case NN_movaps: // Move Aligned Four Packed Single-FP
case NN_movhlps: // Move High to Low Packed Single-FP
case NN_movhps: // Move High Packed Single-FP
case NN_movlhps: // Move Low to High Packed Single-FP
case NN_movlps: // Move Low Packed Single-FP
case NN_movmskps: // Move Mask to Register
case NN_movss: // Move Scalar Single-FP
case NN_movups: // Move Unaligned Four Packed Single-FP
case NN_mulps: // Packed Single-FP Multiply
case NN_mulss: // Scalar Single-FP Multiply
case NN_orps: // Bitwise Logical OR for Single-FP Data
case NN_rcpps: // Packed Single-FP Reciprocal
case NN_rcpss: // Scalar Single-FP Reciprocal
case NN_rsqrtps: // Packed Single-FP Square Root Reciprocal
case NN_rsqrtss: // Scalar Single-FP Square Root Reciprocal
case NN_shufps: // Shuffle Single-FP
case NN_sqrtps: // Packed Single-FP Square Root
case NN_sqrtss: // Scalar Single-FP Square Root
case NN_stmxcsr: // Store Streaming SIMD Extensions Technology Control/Status Register
case NN_subps: // Packed Single-FP Subtract
case NN_subss: // Scalar Single-FP Subtract
case NN_ucomiss: // Scalar Unordered Single-FP Compare and Set EFLAGS
case NN_unpckhps: // Unpack High Packed Single-FP Data
case NN_unpcklps: // Unpack Low Packed Single-FP Data
case NN_xorps: // Bitwise Logical XOR for Single-FP Data
case NN_pavgb: // Packed Average (Byte)
case NN_pavgw: // Packed Average (Word)
case NN_pextrw: // Extract Word
case NN_pinsrw: // Insert Word
case NN_pmaxsw: // Packed Signed Integer Word Maximum
case NN_pmaxub: // Packed Unsigned Integer Byte Maximum
case NN_pminsw: // Packed Signed Integer Word Minimum
case NN_pminub: // Packed Unsigned Integer Byte Minimum
case NN_pmovmskb: // Move Byte Mask to Integer
case NN_pmulhuw: // Packed Multiply High Unsigned
case NN_psadbw: // Packed Sum of Absolute Differences
case NN_pshufw: // Packed Shuffle Word
case NN_maskmovq: // Byte Mask write
case NN_movntps: // Move Aligned Four Packed Single-FP Non Temporal
case NN_movntq: // Move 64 Bits Non Temporal
case NN_prefetcht0: // Prefetch to all cache levels
case NN_prefetcht1: // Prefetch to all cache levels
case NN_prefetcht2: // Prefetch to L2 cache
case NN_prefetchnta: // Prefetch to L1 cache
case NN_sfence: // Store Fence
// Pentium III Pseudo instructions
case NN_cmpeqps: // Packed Single-FP Compare EQ
case NN_cmpltps: // Packed Single-FP Compare LT
case NN_cmpleps: // Packed Single-FP Compare LE
case NN_cmpunordps: // Packed Single-FP Compare UNORD
case NN_cmpneqps: // Packed Single-FP Compare NOT EQ
case NN_cmpnltps: // Packed Single-FP Compare NOT LT
case NN_cmpnleps: // Packed Single-FP Compare NOT LE
case NN_cmpordps: // Packed Single-FP Compare ORDERED
case NN_cmpeqss: // Scalar Single-FP Compare EQ
case NN_cmpltss: // Scalar Single-FP Compare LT
case NN_cmpless: // Scalar Single-FP Compare LE
case NN_cmpunordss: // Scalar Single-FP Compare UNORD
case NN_cmpneqss: // Scalar Single-FP Compare NOT EQ
case NN_cmpnltss: // Scalar Single-FP Compare NOT LT
case NN_cmpnless: // Scalar Single-FP Compare NOT LE
case NN_cmpordss: // Scalar Single-FP Compare ORDERED
// AMD K7 instructions
case NN_pf2iw: // Packed Floating-Point to Integer with Sign Extend
case NN_pfnacc: // Packed Floating-Point Negative Accumulate
case NN_pfpnacc: // Packed Floating-Point Mixed Positive-Negative Accumulate
case NN_pi2fw: // Packed 16-bit Integer to Floating-Point
case NN_pswapd: // Packed Swap Double Word
// Undocumented FP instructions (thanks to norbert.juffa@adm.com)
case NN_fstp1: // Alias of Store Real and Pop
case NN_fcom2: // Alias of Compare Real
case NN_fcomp3: // Alias of Compare Real and Pop
case NN_fxch4: // Alias of Exchange Registers
case NN_fcomp5: // Alias of Compare Real and Pop
case NN_ffreep: // Free Register and Pop
case NN_fxch7: // Alias of Exchange Registers
case NN_fstp8: // Alias of Store Real and Pop
case NN_fstp9: // Alias of Store Real and Pop
// Pentium 4 instructions
case NN_addpd: // Add Packed Double-Precision Floating-Point Values
case NN_addsd: // Add Scalar Double-Precision Floating-Point Values
case NN_andnpd: // Bitwise Logical AND NOT of Packed Double-Precision Floating-Point Values
case NN_andpd: // Bitwise Logical AND of Packed Double-Precision Floating-Point Values
case NN_clflush: // Flush Cache Line
case NN_cmppd: // Compare Packed Double-Precision Floating-Point Values
case NN_cmpsd: // Compare Scalar Double-Precision Floating-Point Values
case NN_comisd: // Compare Scalar Ordered Double-Precision Floating-Point Values and Set EFLAGS
case NN_cvtdq2pd: // Convert Packed Doubleword Integers to Packed Single-Precision Floating-Point Values
case NN_cvtdq2ps: // Convert Packed Doubleword Integers to Packed Double-Precision Floating-Point Values
case NN_cvtpd2dq: // Convert Packed Double-Precision Floating-Point Values to Packed Doubleword Integers
case NN_cvtpd2pi: // Convert Packed Double-Precision Floating-Point Values to Packed Doubleword Integers
case NN_cvtpd2ps: // Convert Packed Double-Precision Floating-Point Values to Packed Single-Precision Floating-Point Values
case NN_cvtpi2pd: // Convert Packed Doubleword Integers to Packed Double-Precision Floating-Point Values
case NN_cvtps2dq: // Convert Packed Single-Precision Floating-Point Values to Packed Doubleword Integers
case NN_cvtps2pd: // Convert Packed Single-Precision Floating-Point Values to Packed Double-Precision Floating-Point Values
case NN_cvtsd2si: // Convert Scalar Double-Precision Floating-Point Value to Doubleword Integer
case NN_cvtsd2ss: // Convert Scalar Double-Precision Floating-Point Value to Scalar Single-Precision Floating-Point Value
case NN_cvtsi2sd: // Convert Doubleword Integer to Scalar Double-Precision Floating-Point Value
case NN_cvtss2sd: // Convert Scalar Single-Precision Floating-Point Value to Scalar Double-Precision Floating-Point Value
case NN_cvttpd2dq: // Convert With Truncation Packed Double-Precision Floating-Point Values to Packed Doubleword Integers
case NN_cvttpd2pi: // Convert with Truncation Packed Double-Precision Floating-Point Values to Packed Doubleword Integers
case NN_cvttps2dq: // Convert With Truncation Packed Single-Precision Floating-Point Values to Packed Doubleword Integers
case NN_cvttsd2si: // Convert with Truncation Scalar Double-Precision Floating-Point Value to Doubleword Integer
case NN_divpd: // Divide Packed Double-Precision Floating-Point Values
case NN_divsd: // Divide Scalar Double-Precision Floating-Point Values
case NN_lfence: // Load Fence
case NN_maskmovdqu: // Store Selected Bytes of Double Quadword
case NN_maxpd: // Return Maximum Packed Double-Precision Floating-Point Values
case NN_maxsd: // Return Maximum Scalar Double-Precision Floating-Point Value
case NN_mfence: // Memory Fence
case NN_minpd: // Return Minimum Packed Double-Precision Floating-Point Values
case NN_minsd: // Return Minimum Scalar Double-Precision Floating-Point Value
case NN_movapd: // Move Aligned Packed Double-Precision Floating-Point Values
case NN_movdq2q: // Move Quadword from XMM to MMX Register
case NN_movdqa: // Move Aligned Double Quadword
case NN_movdqu: // Move Unaligned Double Quadword
case NN_movhpd: // Move High Packed Double-Precision Floating-Point Values
case NN_movlpd: // Move Low Packed Double-Precision Floating-Point Values
case NN_movmskpd: // Extract Packed Double-Precision Floating-Point Sign Mask
case NN_movntdq: // Store Double Quadword Using Non-Temporal Hint
case NN_movnti: // Store Doubleword Using Non-Temporal Hint
case NN_movntpd: // Store Packed Double-Precision Floating-Point Values Using Non-Temporal Hint
case NN_movq2dq: // Move Quadword from MMX to XMM Register
case NN_movsd: // Move Scalar Double-Precision Floating-Point Values
case NN_movupd: // Move Unaligned Packed Double-Precision Floating-Point Values
case NN_mulpd: // Multiply Packed Double-Precision Floating-Point Values
case NN_mulsd: // Multiply Scalar Double-Precision Floating-Point Values
case NN_orpd: // Bitwise Logical OR of Double-Precision Floating-Point Values
case NN_paddq: // Add Packed Quadword Integers
case NN_pause: // Spin Loop Hint
case NN_pmuludq: // Multiply Packed Unsigned Doubleword Integers
case NN_pshufd: // Shuffle Packed Doublewords
case NN_pshufhw: // Shuffle Packed High Words
case NN_pshuflw: // Shuffle Packed Low Words
case NN_pslldq: // Shift Double Quadword Left Logical
case NN_psrldq: // Shift Double Quadword Right Logical
case NN_psubq: // Subtract Packed Quadword Integers
case NN_punpckhqdq: // Unpack High Data
case NN_punpcklqdq: // Unpack Low Data
case NN_shufpd: // Shuffle Packed Double-Precision Floating-Point Values
case NN_sqrtpd: // Compute Square Roots of Packed Double-Precision Floating-Point Values
case NN_sqrtsd: // Compute Square Rootof Scalar Double-Precision Floating-Point Value
case NN_subpd: // Subtract Packed Double-Precision Floating-Point Values
case NN_subsd: // Subtract Scalar Double-Precision Floating-Point Values
case NN_ucomisd: // Unordered Compare Scalar Ordered Double-Precision Floating-Point Values and Set EFLAGS
case NN_unpckhpd: // Unpack and Interleave High Packed Double-Precision Floating-Point Values
case NN_unpcklpd: // Unpack and Interleave Low Packed Double-Precision Floating-Point Values
case NN_xorpd: // Bitwise Logical OR of Double-Precision Floating-Point Values
// AMD syscall/sysret instructions
case NN_syscall: // Low latency system call
case NN_sysret: // Return from system call
// AMD64 instructions
case NN_swapgs: // Exchange GS base with KernelGSBase MSR
// New Pentium instructions (SSE3)
case NN_movddup: // Move One Double-FP and Duplicate
case NN_movshdup: // Move Packed Single-FP High and Duplicate
case NN_movsldup: // Move Packed Single-FP Low and Duplicate
// Missing AMD64 instructions
case NN_movsxd: // Move with Sign-Extend Doubleword
case NN_cmpxchg16b: // Compare and Exchange 16 Bytes
// SSE3 instructions
case NN_addsubpd: // Add /Sub packed DP FP numbers
case NN_addsubps: // Add /Sub packed SP FP numbers
case NN_haddpd: // Add horizontally packed DP FP numbers
case NN_haddps: // Add horizontally packed SP FP numbers
case NN_hsubpd: // Sub horizontally packed DP FP numbers
case NN_hsubps: // Sub horizontally packed SP FP numbers
case NN_monitor: // Set up a linear address range to be monitored by hardware
case NN_mwait: // Wait until write-back store performed within the range specified by the MONITOR instruction
case NN_fisttp: // Store ST in intXX (chop) and pop
case NN_lddqu: // Load unaligned integer 128-bit
// SSSE3 instructions
case NN_psignb: // Packed SIGN Byte
case NN_psignw: // Packed SIGN Word
case NN_psignd: // Packed SIGN Doubleword
case NN_pshufb: // Packed Shuffle Bytes
case NN_pmulhrsw: // Packed Multiply High with Round and Scale
case NN_pmaddubsw: // Multiply and Add Packed Signed and Unsigned Bytes
case NN_phsubsw: // Packed Horizontal Subtract and Saturate
case NN_phaddsw: // Packed Horizontal Add and Saturate
case NN_phaddw: // Packed Horizontal Add Word
case NN_phaddd: // Packed Horizontal Add Doubleword
case NN_phsubw: // Packed Horizontal Subtract Word
case NN_phsubd: // Packed Horizontal Subtract Doubleword
case NN_palignr: // Packed Align Right
case NN_pabsb: // Packed Absolute Value Byte
case NN_pabsw: // Packed Absolute Value Word
case NN_pabsd: // Packed Absolute Value Doubleword
// VMX instructions
case NN_vmcall: // Call to VM Monitor
case NN_vmclear: // Clear Virtual Machine Control Structure
case NN_vmlaunch: // Launch Virtual Machine
case NN_vmresume: // Resume Virtual Machine
case NN_vmptrld: // Load Pointer to Virtual Machine Control Structure
case NN_vmptrst: // Store Pointer to Virtual Machine Control Structure
case NN_vmread: // Read Field from Virtual Machine Control Structure
case NN_vmwrite: // Write Field from Virtual Machine Control Structure
case NN_vmxoff: // Leave VMX Operation
case NN_vmxon: // Enter VMX Operation
return false;
break;
default:
msg("ERROR: Unknown instruction opcode at %x : %s\n", this->GetAddr(),
this->GetDisasm());
break;
} // end switch on opcode
return true;
} // end SMPInstr::BuildRTL()
// Iterate through all reg transfers and call SyncRTLDefUse for each.
void SMPInstr::SyncAllRTs(void) {
for (size_t index = 0; index < this->RTL.GetCount(); ++index) {
this->SyncRTLDefUse(this->RTL.GetRT(index));
}
return;
} // end of SMPInstr:SyncAllRTs()
// Ensure that each operand of the RTL is found in the appropriate DEF or USE list.
void SMPInstr::SyncRTLDefUse(SMPRegTransfer *CurrRT) {
// The Guard expression and ExtraKills are almost never represented in the DEF and USE
// lists. When they are, they are added in MDFixupDefUseLists(), so we ignore them here.
// The only DEFs should come from left operands of SMP_ASSIGN operators, i.e. the effects
// of register transfers.
op_t LeftOp, RightOp;
set<DefOrUse, LessDefUse>::iterator CurrDef, CurrUse;
bool DebugFlag = false;
#if SMP_VERBOSE_DEBUG_BUILD_RTL
DebugFlag |= (0 == strcmp("__libc_csu_fini", this->BasicBlock->GetFunc()->GetFuncName()));
#endif
if (DebugFlag) {
msg("SyncRTLDefUse entered. Dump of USE list:\n");
this->Uses.Dump();
}
LeftOp = CurrRT->GetLeftOperand();
if (SMP_ASSIGN == CurrRT->GetOperator()) {
assert(o_void != LeftOp.type);
assert(o_imm != LeftOp.type);
CurrDef = this->Defs.FindRef(LeftOp);
if (CurrDef == this->GetLastDef() && !LeftOp.is_reg(R_ip)) {
#if SMP_VERBOSE_DEBUG_BUILD_RTL
msg("WARNING: DEF not found for SMP_ASSIGN in %s ; added op:", this->GetDisasm());
PrintOperand(LeftOp);
msg("\n");
#endif
this->Defs.SetRef(LeftOp, CurrRT->GetOperatorType());
}
}
else { // not SMP_ASSIGN; left operand should be a USE
if (o_void != LeftOp.type) {
CurrUse = this->Uses.FindRef(LeftOp);
if (CurrUse == this->GetLastUse()) {
#if SMP_VERBOSE_DEBUG_BUILD_RTL_DEF_USE
msg("WARNING: USE not found for ");
PrintOperand(LeftOp);
msg(" in %s ; added\n", this->GetDisasm());
#endif
this->Uses.SetRef(LeftOp);
}
}
}
if (!CurrRT->HasRightSubTree()) {
RightOp = CurrRT->GetRightOperand(); // right operand should be a USE
if (o_void != RightOp.type) {
CurrUse = this->Uses.FindRef(RightOp);
if (CurrUse == this->GetLastUse()) {
#if SMP_VERBOSE_DEBUG_BUILD_RTL_DEF_USE
msg("WARNING: USE not found for ");
PrintOperand(RightOp);
msg(" in %s ; added\n", this->GetDisasm());
#endif